The effects of fire and harvesting on Restionaceae SPP. (Thamnochortus insignis and T. erectus) with different life histories : a matrix modelling approach

Hele tekst

(1)THE EFFECTS OF FIRE AND HARVESTING ON RESTIONACEAE SPP. (THAMNOCHORTUS INSIGNIS AND T. ERECTUS) WITH DIFFERENT LIFE HISTORIES: A MATRIX MODELLING APPROACH. TESSA ANGELA CAMPBELL. Thesis presented in partial fulfilment of the requirements for the degree of Master of Science at the University of Stellenbosch. Supervisor: Prof. K.J. Esler Co-supervisor: Prof. D. Ward December 2006.

(2) DECLARATION. I, the undersigned, hereby declare that the work contained in this thesis is my own original work and that I have not previously in its entirety or in part submitted it at any university for a degree.. Signature:……………………….. Date:…………………………... i.

(3) Summary. The Restionaceae is a dominant family in the Fynbos Biome, an area in which fire plays a role as an important disturbance, yet little is known about their population dynamics. Two species of the Restionaceae (Thamnochortus insignis and T. erectus) are economically important as thatching reed and differ in their life-histories. This study aims to determine the effects of variation in life history (sprouter vs. non-sprouter) on the population structure and dynamics of T. erectus (“wyfies riet”, sprouter) and T. insignis (“mannetjies riet”), a non-sprouting species.. A matrix-modelling approach based on field data. collected by Ball (1995) is used to determine population growth rates, stable stage distributions and stage sensitivity and elasticity for the two species with no disturbance present. The sprouter (T. erectus) shows a positive population growth rate (λ >1) and greater persistence within all stages. The non-sprouting species (T. insignis) shows a negative population growth rate (λ <1) between disturbances as well as greater seed production, germination and growth between stages. Based on the population dynamics of these two species, further research was done to understand the effect of disturbance (harvesting and fire) on these species. A matrix modelling approach was used to determine which disturbance frequency maximises population output and success. Harvesting as well as fire results in a decline in T. insignis populations. A five year frequency for harvesting results in the greatest output of adult plants with the lowest effect on the population, and a fire frequency of 50 to 65 years is recommended. Testing indicates that the model underestimates the number of adults in the population and thus the model is conservative. T. erectus populations grow despite fire or harvesting; thus any reasonable harvesting (3-5 year frequency) and fire (10+ years between fire) regime would ensure population persistence. As data were limited it was not possible to test the. ii.

(4) results although T. erectus appears resilient to disturbance and therefore a predetermined regime is not as important as in T. insignis. Recommendations to farmers are made based on these results.. Opsomming Die Restionaceae is ‘n funktionele tipe van die Fynbos Bioom, ‘n gebied waar veldbrande ‘n belangrike rol speel as ‘n versteuring. Daar is alhoewel min kennis oor hul populasie dinamika. Twee spesies van die Restionaceae (Thamnochortus insignis en T. erectus) is ekonomies baie belangrik vir dekriet en verskil in hul lewens-geskiedenis. Hierdie werkstuk se doel is om die impak van die variasie in die lewens-geskiedenis (spruitendende spesie tn. nie-spruitende spesie) op die populasie struktuur en dinamika van T. erectus (mannetjies riet – spruitendende spesie) en T. insignis (wyfies riet – niespruitende spesie) te bepaal. ‘n Matriks-model gebaseer op veld, versamelde data deur Ball (1995) is gebruik om die populasie groeikoers, stabiele vlak van verspreiding en sensitiviteit en elastisiteit te bepaal, vir die twee spesies sonder enige huidige versteurings. Die spruit spesie (T. erectus) wys ‘n positiewe populasie groeikoers (λ >1) en grooter weerstand op alle vlakke. Die nie-spruitende spesie (T. insignis) wys ‘n negatiewe populasie groeikoers (λ <1) tussen versteurings asook grooter saad produksie, ontkieming en groei tussen vlakke as die spruit species. Gebaseer op die populasie dinamika van die twee spesies verdere studies was gedoen om die effek van versteurings (oes en veldbrande) op die spesies te bepaal. ‘n Matriks model benadering was gebruik om te bepaal watter tydperke tussen versteurings die populasie opbrengs en sukses maksimiseer. Gevolglik verminder oes die T .insignis se populasie sowel as veldbrande. ‘n Vyf jaar frekwensie vir oes veroorsaak die grootste opbrengs van volwasse plante met die laagste effek op die populasie, en ‘n veldbrand elke 50 tot 65 jaar word aanbeveel. Toetse wys dat die model. iii.

(5) die getal volwassenes in die populasie onderskat en dus is die model konserwatief. T. erectus populasies groei ten spyte van oes of veld brande. Dus sal enige redelike oes en veldbrand program ‘n weerstand in populasie groei waarborg. As gevolg van beperkte data, was dit nie moontlik om die resultate te toets nie, maar dit lyk wel asof T. erectus weerstandig is teenoor versteurings en dus is ‘n vooruitbepaalde program nie so belangrik soos in T. insignis nie. Aanbevelings vir boere is saamgestel gebaseer op hierdie resultate.. iv.

(6) Contents:. Declaration ...................................................................................................................ii Summary......................................................................................................................iii Opsomming .................................................................................................................iv Contents .......................................................................................................................vi List of figures ..............................................................................................................ix List of tables ................................................................................................................xi Acknowledgements ...................................................................................................xiii. Chapter One: Introduction ............................................................................................1 1.1. Introduction ............................................................................................... 1 1.2. Rationale....................................................................................................2 1.3. Literature review: Construction, benefits and testing of matrix models ... 4 1.3.1. Matrix model construction ............................................................... 4 1.3.2. Matrix analysis .................................................................................9 1.3.3. The benefits of using models.......................................................... 12 1.3.4. Model testing ..................................................................................15 1.4. Literature review: The Fynbos Biome..................................................... 16 1.5. Literature review: Economic uses of Fynbos .......................................... 18 1.6. Literature review: Fire ............................................................................. 21 1.6.1. Fire and life-histories...................................................................... 21 1.6.2. Components of Fire ........................................................................23 1.6.3. Non-sprouter: Sprouter balance......................................................24 1.7. Literature review: Life-history strategies ................................................ 26. v.

(7) 1.8. Literature review: Restionaceae .............................................................. 31 1.8.1. Reproductive Biology.....................................................................31 1.8.2. General Ecology ............................................................................. 33 1.8.2.1. Water and nutrient relations .................................................. 33 1.8.2.2. Growth form ..........................................................................34 1.8.2.3. Fire Survival ..........................................................................35 1.8.2.4. Grazing ..................................................................................35 1.8.3. Genus Thamnochortus....................................................................36 1.9. Literature Review: The thatching industry.............................................. 47 1.9.1. Economic uses of Restionaceae......................................................47 1.9.2. Thatching material..........................................................................48 1.9.3. Thatched roofs ................................................................................48 1.9.4. Harvesting of Thamnochortus insignis and T. erectus for thatching reed ................................................................................................. 49 1.10. Key questions.......................................................................................55 1.11. References............................................................................................56 Chapter Two: Contrasting life history strategies in Restionaceae species (Thamnochortus insignis and T. erectus): A matrix modelling approach .................. 67 2.1. Abstract....................................................................................................67 2.2. Introduction ............................................................................................. 68 2.2.1. Biology of the Restionaceae ......................................................... 68 2.2.2. Species description........................................................................69 2.2.3. Economic uses............................................................................... 70 2.2.4. Life-history theory ........................................................................ 70 2.2.5. Matrix models ............................................................................... 73. vi.

(8) 2.2.6 Hypotheses and key questions ....................................................... 74 2.3. Methods ...................................................................................................75 2.3.1. Parameter estimation..................................................................... 75 2.3.2. Model construction .......................................................................79 2.4. Results ..................................................................................................... 84 2.4.1. Parameters..................................................................................... 84 2.4.2. Population projection and stable stage distribution ...................... 84 2.4.3. Sensitivity and elasticity matrices................................................. 86 2.5. Discussion................................................................................................93 2.6. References ...............................................................................................96. Chapter Three: The effects of fire and harvesting on Restionaceae species (Thamnochortus insignis and T. erectus) utilised in the thatching industry......................................... 99 3.1. Abstract....................................................................................................99 3.2. Introduction ........................................................................................... 100 3.2.1. Fire in the Fynbos Biome............................................................ 100 3.2.2. Restionaceae................................................................................ 101 3.2.3. The Thatching Industry and harvesting of Thamnochortus insignis and T. erectus ..............................................................................102 3.2.4. Matrix models and Analysis ....................................................... 103 3.2.5. Key Questions and hypotheses ...................................................105 3.3. Method...................................................................................................106 3.3.1 Parameter estimation.................................................................... 106 3.3.2. Model construction .....................................................................108. vii.

(9) 3.3.3. Model testing...............................................................................110 3.4. Results ................................................................................................... 116 3.4.1. Effects of harvesting ...................................................................116 3.4.2. Effects of fire ..............................................................................118 3.4.3. Model testing...............................................................................119 3.5. Discussion..............................................................................................126 3.5.1. The effects of harvesting on Thamnochortus insignis and T. erectus...............................................................................128 3.5.2. The effects of fire on Thamnochortus insignis and T. erectus ..............................................................................129 3.5.3 Model testing................................................................................ 131 3.5.4. Conclusion ..................................................................................132 3.6. References .............................................................................................133. Chapter Four: Conclusions and recommendations ..............................................136 4.1. Introduction ........................................................................................... 136 4.2. Results and conclusions.........................................................................136 4.3. Suggestions for farmers ......................................................................... 139 4.4. Limitations and recommendations for future study............................... 140 4.5. References .............................................................................................141 Appendix One ......................................................................................................142 Appendix Two......................................................................................................145. viii.

(10) List of figures. Fig. 1.1. (a) A life cycle graph indicating the parameters used in the model and (b) the incorporation of these parameters into the matrix. ................................... 8 Fig. 1.2. a) The female flower, b) male flower and c) seedling of Thamnochortus insignis................................................................................................................. 43 Fig. 1.3. a) The female flower, b) male flower and c) seedling of Thamnochortus erectus. ................................................................................................................ 44 Fig. 1.4. a) The distribution of Thamnochortus insignis and b) T. erectus ....................... 45 Fig. 1.5 a) The growth and flowering of Thamnochortus insignis and b) T. erectus over the seasons........................................................................................................... 46 Fig. 1.6. A farm cottage in the Bredasdorp region thatched with Thamnochortus insignis53 Fig. 1.7. Recently cut Thamnochortus insignis population near Albertinia ...................... 53 Fig. 1.8. Bundles of thatch standing in ricks near Albertinia waiting to be loaded for transport............................................................................................................... 54 Fig. 1.9. Bundles of thatch being loaded onto a truck for transport .................................. 54 Fig. 2.1. Total population growth including density dependence and population growth rate of T. insignis (a) and T. erectus (b) over 100 years without disturbance ..... 88 Fig. 2.2. The stable stage distribution of T. insignis and T. erectus as log percentage of the total population.............................................................................................. 89 Fig. 2.3. The stable stage distribution of T. erectus excluding and including density dependence as log percentage of the total population......................................... 90 Fig. 2.4. Triangular ordination representing corresponding elasticity values for stasis (S), growth (G) and fecundity (F) of T. insignis and T. erectus ......................... 92. ix.

(11) Fig. 3.1. The population growth rate and population size of above ground individuals (plotted on a logarithmic scale) in T. insignis (a) and T. erectus (b) plotted on a logarithmic scale over 100 years with a harvesting frequency of 8 years with the first harvest occurring in year 8.......................................................................... 120 Fig. 3.2. The population growth rate and population size of above-ground individuals in T. insignis (a) and T. erectus (b) plotted on a logarithmic scale over 100 years with a fire frequency of 25 years with the first fire occurring in year 25 ......... 122. x.

(12) List of Tables. Table 1.1. Properties of Thamnochortus insignis and Thamnochortus erectus from Linder (1990b). .............................................................................................. 42 Table 2.1. Summarised data from Ball (1995) used in the estimation of population parameters for T. insignis and T. erectus........................................................ 82 Table 2.2. A transition matrix of probabilities for T. insignis (a) and T. erectus without (b) and with density dependence included (c) ................................................ 83 Table 2.3. Sensitivity values of (a) T. insignis and (b) T. erectus ................................... 91 Table 2.4. Elasticity values of (a) T. insignis and (b) T. erectus ..................................... 91 Table 3.1. Summarised data from Ball (1995) used in the estimation of population parameters for harvested vegetation in T. insignis and T. erectus................ 112 Table 3.2. Summarised data from Ball (1995) used in the estimation of population parameters for fire-disturbed vegetation in T. insignis and T. erectus ......... 113 Table 3.3. A transition matrix of probabilities for T. insignis during harvesting (a), fire (b) and with no disturbance (c)..................................................................... 114 Table 3.4. A transition matrix of probabilities for T. erectus during harvesting (a), fire (b), and with no disturbance (c) and including density dependence (d).................. 115 Table 3.5. The total number of adult individuals harvested over 100 years in T. insignis (a) and T. erectus (b) at different frequencies from once every 8 years.............................................................................................................. 121 Table 3.6. The total number of adult individuals present at the time of each fire summed over 100 years in T. insignis (a) and T. erectus (b) at different fire frequencies.................................................................................................... 123. xi.

(13) Table 3.7. Observed field data means and standard deviations and expected data calculated as a stage distribution from the model predictions under the same disturbance history (time since harvest) for T. insignis................................ 124 Table 3.8. χ2 goodness of fit test of observed field data versus expected data based on model predictions of stage distribution for T. insignis ................................. 125. xii.

(14) Acknowledgements. Thanks go to my supervisors, Prof. K. Esler and Prof. D. Ward for support and encouragement. I’m grateful for the inputs and comments of many people, specifically Stuart Bell, Jessica Conradie, Marie-Inez Botha and Dr. K. Stewart. Thanks to Carol Horvitz for the algorithm for sensitivities and elasticities. Thanks also go to Werner Gerber for assisting me with my field work and sharing the experiences. To the Janse van Rensburg family of Albertinia for the welcome into their home and their willingness to share their beautiful farm with me, as well as to the other farm owners in the Bredasdorp region who allowed me to enter their properties. To the managers of De Hoop nature reserve for allowing me access to restricted areas. Thanks also go to the NRF for supporting this study with a post-graduate bursary.. A personal word of thanks to my parents. To my mother, Ann, for helping me over the last hurdle. To my father, James, for keeping an eye on the future when I was looking at my toes, this one is for us both.. xiii.

(15) Chapter One: General Introduction. 1.1. Introduction. This study aimed to develop a sustainable harvest and fire regime for Thamnochortus insignis and T. erectus (Restionaceae). These species are used as thatching reed and occur in the Fynbos Biome (Linder, 1991). The Fynbos Biome occurs in the southwestern part of South Africa, has winter rainfall and is characterised by the presence of Proteaceae, Ericaceae and Restionaceae (Cowling et al., 1997). Fire is an important disturbance in Fynbos systems and a variety of life history strategies have developed in response to variance in fire history (Bond and van Wilgen, 1995).. Thamnochortus insignis relies on seeds for population. regeneration and thus is a non-sprouter. Thamnochortus erectus relies on resprouting and on seeds for population persistence and thus is a sprouter. This study aimed to determine how population dynamics were influenced by life history strategies using a matrix modelling approach. Matrix models enabled projection of stage classes in a population (Caswell, 2001) and thus a management strategy that maximises long-term population output could be determined.. The accuracy and applicability of projected results were tested by using. independent field collected data from natural populations.. This thesis is structured so that each data chapter (Chapters 2 & 3) is written as a stand alone research paper for submission to Biological Conservation. As a result there was some methodological repetition in the thesis, although an attempt was made to keep this to a minimum. The first chapter is an introduction and review of the available literature. The second chapter discusses life history strategies and makes predictions of population growth under different management scenarios.. The third chapter deals with the management. 1.

(16) strategies and the development of a sustainable approach to harvesting that maximises longterm output. Further, outputs of the model are tested against field-collected data. Chapter four is a conclusion and attempts to make recommendations for managers or farmers.. 1.2. Rationale. Sustainability is a keyword when referring to the use of ecological resources.. A. sustainable system is defined as one that survives or persists (Costanza and Patten, 1995) both temporarily and in the long run over its full, expected life span. To biologists this means avoiding extinction and living to survive and reproduce, while economically it means avoiding major disruptions and collapses in the resource in the long term (Costanza et al., 1997). The promotion of sustainable use, and specifically the ‘improvement of harvesting practices in both terrestrial and marine environments’ is a goal of the Cape Action Plan for People and the Environment (CAPE) (Gelderblom et al., 2003). Projects proposed by CAPE in an attempt to improve the efficiency and sustainability of harvesting include models for harvesting wildflowers (Gelderblom et al., 2003).. Thatching reed (Thamnochortus insignis) is economically important as a resource with a potential for export overseas (Linder, 1990a), a closely related congeneric species, T. erectus, is also used as thatching material (Anon, 1998). The advantages of thatch increase its appeal as a harvestable resource; these include versatility, insulating properties for temperature and sound, and simple and inexpensive maintenance (Long, 1978). Thatching reed is generally a locally produced material, which means that supplies can be increased at a low cost and without upsetting the local economy (Hall, 1988). It also does not require sophisticated machinery or tools and hence is suitable as the basis of a small business with benefits to. 2.

(17) farmers and the local community (Hall, 1988). Not only is it low cost farming, it is also a potentially fruitful resource due to the fact that city dwellers spend twice as much on thatch as on any other roofing material (Davis, 1993). Disadvantages of using thatch include greater fire risk, susceptibility to decay and decomposition and the possible provision of refuge to vermin (Anon, 1998). There is great potential for the cultivation of these plants should the market expand (Van Wilgen et al., 1992).. The natural vegetation of the Agulhas Plain, where both species occur, has the potential to generate income in an economically and ecologically sustainable way; two of the most important industries using natural vegetation are wild flower harvesting and ecotourism (Heydenrych, 1996). The duneveld of the southern Western Cape is not good agricultural land and productivity is limited by variable rainfall and infertile soils (Davis, 1993). The wildflower industry (including the thatching industry) has increased the value of marginal agricultural land (Van Wilgen et al., 1992) and thus thatching reed provides a supplementary income (Davis, 1993). The thatching industry, from Thamnochortus species, has an annual value of between US$1.3 million and US$1.9 million (exchange rate of R7.50 = US$ 1.00) (Turpie et al., 2003).. Based on the economic importance of the species, a harvesting strategy that results in sustainable use of the resource is necessary (Cowling et al., 1997). The economic importance of the species indirectly acts as a draw card for conservation, as quantification and appreciation by society of the services provided by a natural ecosystem provides a powerful incentive for their conservation (Cowling and Costanza, 1997). A harvesting strategy that gives further insight into the population dynamics of the two thatching species is thus important regarding economics and conservation. This study used matrix models to project. 3.

(18) population growth, model population dynamics and model the effects of disturbance (e.g. fire and harvesting) on the dynamics. More research is needed to make recommendations for sustainable harvesting of plants with soil-stored seed banks, such as T. insignis and T. erectus (Cowling et al., 1997). An additional benefit of the study was the comparison between the two species with two different life histories and their varying responses to certain disturbance factors, namely harvesting and fire. Very little research has been done on the biology of sprouters in fynbos and it is important to understand their role in ecosystem processes (Cowling et al., 1997). Thatch production is a key factor in the development of any long-term plan on the duneveld, and science must contribute to the information on which any plan is based (Davis, 1993).. 1.3. Literature Review: Construction, benefits and testing of matrix models. 1.3.1. Matrix model construction. The method used for population projection and analysis of population dynamics in this study was transition matrix modelling. The demographic parameters of a population are determined by the vital rates of each individual in the population (Caswell, 2001). The vital rates include individual birth, growth, maturation, fertility and mortality (Caswell, 2001). Generally the transition rates (growth and persistence) are between zero and one with the exception being asexual vegetative growth which could result in a transition rate of greater than one as each individual produces more individuals. Fecundity is generally greater or equal than zero except when seed predation and loss from a seed bank are built into this value. Matrix population models provide a link between the vital rates of individuals and the population dynamics based on a simple description of the life cycle (Caswell, 2001). The life. 4.

(19) cycle is simplified to biologically meaningful classes such as age. The contribution an average individual in a class (say j) makes in a time interval (t to t+1) to another class (say i) is expressed as a coefficient (aij) of a square matrix A (Silvertown et al., 1993). The matrix (A) is square with the number of rows and columns being equal to the number of classes (Silvertown et al., 1993).. Matrix population models provide a means of calculating the population growth rate and assessing the influence of each vital rate on population growth rate (Caswell, 2001). Deterministic single-population models (this study) are useful as they are simple and require the least amount of data in comparison to stochastic single-population modelsincorporate chance events by ing, metapopulation models and spatially explicit models (Beissinger and Westphal, 1998). In deterministic single-population models, the required data is limited to: stage of first reproduction, reproductive success estimates and survivorship; these demographic rates remain constant over time, which simplifies the model (Beissinger and Westphal, 1998). Analytical models (mathematical models whose solution is obtained purely by mathematical argument (Haefner, 1996)) are useful in determining system behaviour rather than making predictions and testing model assumptions but have a limited use in management (Beissinger and Westphal, 1998). Stochastic models require more than twice the amount of data needed for deterministic models because variance in fecundity and survival as well as carrying capacity are included in the models (Beissinger and Westphal, 1998). They include the effects of management options but do not have a spatial component like metapopulation models and spatially explicit models. Metapopulation models deal with migration between habitat patches, while spatially explicit models include landscape processes and are grid based and thus require immense data (Beissinger and Westphal, 1998). Deterministic single-. 5.

(20) population models are most appropriate for this study as they provide management answers from limited data.. Early matrix models were age-classified matrix models (Leslie, 1945, 1948) but Lefkovitch (1965) introduced the concept of stage-classified matrix models. Plant ecologists have adopted these models because size is often a better predictor of plant performance than age (Caswell, 2001), as seen for example in Harper and White (1974), Werner and Caswell (1977) and Morris and Doak (1998). Lefkovitch matrix models also allow transitions from later stages to earlier stages (retrogression) (O’Connor, 1993). It is assumed that individuals in the same stage of a lefkovitch matrix experience the same probabilities of mortality, fecundity and growth (Caswell, 2001). These stages need to be biologically meaningful and capture the relevant life history of the plant (Caswell, 2001). Thamnochortus insignis and T. erectus exhibited four stages. The four stages were: seeds in the soil-stored seed bank, seedlings, non-reproductive adults and reproductive adults (Fig 1.1).. Lefkovitch matrix models use the general form: A x n (t) = n (t + 1). (1). where A is the population matrix, n (t) is the population vector of a single column of numbers representing the current population stage-class distribution and n (t + 1) is the expected population stage-class distribution (Caswell, 2001). The matrix A is known as the population projection or transition matrix and contains the mean probabilities that describe the proportion of individuals that survive and remain within a stage (persistence), those that enter another stage (growth or retrogression) and the production of offspring into the first stage (fecundity) (Fig. 1.1) (Caswell, 2001).. The matrix characterises the nature of the population by. 6.

(21) ncorporating mean fecundity, mortality and growth rates for each stage in the life cycle (Caswell, 2001).. 7.

(22) a) FR. Seed bank. GSB. Seedlings. GS. Non-reproductive GNR adults. RGR. PSB. PR. PNR. PS. Reproductive adults. b). SB. S. NR. R. SB. PSB. 0. 0. FR. S. GSB. PS. 0. 0. NR. 0. GS. PNR. RGR. R. 0. 0. GNR. PR. Fig. 1.1. (a) A life cycle graph indicating the parameters used in the model and (b) the incorporation of these parameters into the matrix.. Growth to the next stage (G) and. persistence within a stage (P) occurred in the seed bank (SB), seedlings (S) and nonreproductive adult stages (NR). Retrogression to a previous stage (RG), fecundity (F) and persistence within a stage (P) occurred in the reproductive stage (R). The columns of the matrix are the stages at time t and the rows the stages at time t+1.. 8.

(23) 1.3.2. Matrix analysis. Matrix models have advantages and disadvantages; reality is sacrificed to gain the advantage of mathematical properties of the formulation that allow the modeller to examine consequences without experimentation on the model (Thalen et al., 1987). A population can be projected into the future in a constant environment (Caswell, 2001) and the effects of a disturbance on the population can be simulated using a simulation model. Matrix projection assumes unchanging conditions, which lead to unrealistic long-term predictions with exponential population growth (Menges, 1998). A disturbance affects the vital rates of the population and these changes determine the population’s response. A matrix population model that includes these vital rates will project the effects of disturbance on the population. The projected stage distribution of the population will approach stability such that the proportion of individuals in each stage becomes constant (Caswell, 2001). This is known as the stable-stage distribution (SSD) and is the dominant right-hand eigenvector of the transition matrix (Caswell, 2001). When SSD is reached, the population growth rate or rate of population change is stable. This rate is determined as:. λ = n (t) / (n (t -1)). (2). and is the dominant eigenvalue of the matrix (Caswell, 2001). The time required for stability to occur depends on how similar the initial stage distribution is to the SSD (Caswell, 2001). The population growth rate (λ) has a value of 1.0 when the total population remains constant through time i.e. the population is neither increasing nor decreasing (Caswell, 2001). The population growth rate is greater than 1.0 when the population is increasing and less than 1.0 when the population is declining (Caswell, 2001). Both the SSD and λ are functions of. 9.

(24) the vital rates and will remain constant provided the transition matrix is unchanged (Caswell, 2001).. Matrix analyses have been developed to determine the effect of a change in vital rates (Caswell, 2001). These are known as perturbation analyses and ask what would happen to a dependent variable if one or more independent variables were to change (Caswell, 2001). The uses include predicting the results of future changes in the vital rates (e.g. due to fire or harvesting) and designing sampling schemes because information on the sensitivity of the vital rates can highlight the vital rates where accuracy matters most (Caswell, 2001). Sensitivity analysis is the calculation of the change in λ because of a change in a vital rate or matrix element (aij) (Caswell, 2001). A sensitivity matrix of λ to all the matrix elements can be constructed using: Sij = δ λ / δaij. (3). Sensitivity analysis gives the effect on λ of changes in any entry of the matrix including those that are fixed at zero (i.e. growth directly from seed to reproductive adult) (Caswell, 2001). The derivative (S) indicates what would happen to λ if the element (aij) were to change, even if such change is a biologically impossible event (Caswell, 2001). This can answer theoretical questions and is not always irrelevant (Caswell, 2001).. Transition probabilities (which are between zero and one) and fertilities (greater than or equal to zero) are measured on different scales (Caswell, 2001). Proportional, rather than absolute, perturbations would allow comparison between different scales (De Kroon et al., 1986). Sensitivity also does not measure the contribution of an element to λ (De Kroon et al., 1986). Caswell et al. (1984) and De Kroon et al. (1986) developed a measure of proportional sensitivity called elasticity. The method corresponds with other previously used methods (De. 10.

(25) Kroon et al., 1986). The elasticity is calculated as the proportional change in λ resulting from a proportional change in a matrix element (De Kroon et al., 1986):. eij = (aij / λ) (δ λ / δaij).. (4). The elasticity (e) quantifies the contribution of the matrix element to λ and all elasticities sum to one (De Kroon et al., 1986). Elasticity measures proportional sensitivity and describes the degree to which population growth is determined by transitions (De Kroon et al., 1986). Sensitivity predicts the impacts of hypothetical changes in parameters on the population growth rate while elasticity compares the potential shifts with real life history patterns (De Kroon et al., 1986).. Elasticity analysis has become a commonplace tool in ecology, particularly in the management of threatened species (Benton and Grant, 1999). Silvertown et al. (1996) applied elasticity matrices to plant population conservation. A previous study (Silvertown et al., 1993) compared 21 species of woody plants and 45 herbaceous perennials (and 15 species were added in Silvertown et al. (1996)), with a range of life histories and habitats, using elasticity analysis. In both studies, elasticities within regions of the matrix were summed (Silvertown et al., 1996). All transitions from a stage to a later stage were summed to give a value for G, which measures the effects of individual growth and clonal growth (Silvertown et al., 1996). The L value measures stasis and the measure of individuals staying in the same class as well as retrogression or transition to a previous stage (Silvertown et al., 1996). The F value is the fecundity (the recruitment of seeds or seedlings) (Silvertown et al., 1996). The G/L/F ratios of the species were plotted on a triangle using triangular ordination and clear relationships between the G/L/F ratios and the functional groups were observed (Silvertown et al., 1996). Functional groups occur in characteristic regions of the triangle (Silvertown et al.,. 11.

(26) 1996). The elasticity values are high for growth and fecundity in semelparous herbs, for survival and fecundity in forest herbs, for growth and survival in trees while those of iteroparous herbs of open habitats and shrubs are less clear (Silvertown et al., 1996). Silvertown et al. (1996) also found that variation occurs between populations but single population studies remain useful because the populations are part of a successional trajectory (Silvertown et al., 1996). Elasticity analysis indicates life history stages that sustain growing populations but it may mislead management if λ is less than one (Silvertown et al., 1996).. Elasticity analysis can be applied to conservation, bio-control and sustainability but it is important to use both sensitivity and elasticity analyses wisely (Benton and Grant, 1999). Elasticity and sensitivity analyses are important tools in population ecology because of their ease of interpretation and analytical simplicity (Benton and Grant, 1999).. However,. elasticities are calculated from a density independent, time invariant model and it is unknown whether the elasticities can predict real population response after management (Benton and Grant, 1999). Further, elasticties are constrained by the matrix model (Benton and Grant, 1999). A model with fewer and broader classes will emphasize stasis rather than growth (Benton and Grant, 1999). Elasticity analysis remains a useful tool but care must be taken in interpretation (Benton and Grant, 1999). Matrix analysis is not limited to elasticity and sensitivity analysis (prospective analyses); other analyses include retrospective perturbation analyses (Caswell, 2000) and second derivatives (Caswell, 1996).. 1.3.3. The benefits of using models. A model is a description of a system (population), which is a collection of interrelated objects (individual plants), which are elemental units upon which observations can be made. 12.

(27) (Haefner, 1996). Models have three main uses in science: to understand, to predict and to control (Haefner, 1996). These uses influence the structure of the model with trade-offs between the uses and properties of the model (Haefner, 1996). Models vary in their realism (the structure of the model mimics the real world), precision (accuracy of the model output) and generality (the number of systems that the model can be applied to) (Haefner, 1996). It is important to consider these model uses and properties when criticising models but models can be misapplied when there is uncertainty in the data or the underlying mechanisms are not understood (Haefner, 1996).. Criticism has been directed at models; the common misconceptions are about gaps in understanding and data, model validation, the realism of models, predictive uses of models as well as the time and cost involved (Starfield, 1997). A model must be measured based in how well it meets its purpose and whether it meets its purpose better than any other tool (Starfield, 1997). All criticism can be countered under this standard of measurement. It is not necessary to have complete understanding of the system or complete data to build a model (Starfield, 1997). The purpose of the model will also endorse the use of a simple model; simple models are preferential if more easily understood (Starfield, 1997). Models are essentially problemsolving tools that do not attempt to predict the future; it is not necessary that they are large and have multiple purposes (Starfield, 1997). A good model will enable comparison of model results to a real system (Hannon and Ruth, 1997). Model components can be manipulated to observe how these components affect the rest of the system (Hannon and Ruth, 1997). Models can be predictive, forecasting the future course of a system as well as highlighting gaps in knowledge and indicating normal fluctuations (Hannon and Ruth, 1997). Modelling is a useful tool as it orders thought and provides a deeper understanding of the system (Hannon and Ruth, 1997).. 13.

(28) The benefits of using models have led to their application in decision-making and understanding of natural systems.. Matrix models are also often used as tools to aid. understanding due to the available analyses that accompany the models.. Models are. increasingly used to determine the probability of population survival in endangered plants (Groom and Pascual, 1998) where experimentation is risky due to the limited number of individuals, this is known as population viability analysis (PVA).. PVA’s analyse the. probability of extinction and a minimum viable population can be calculated (Menges, 1998). Beissenger and Westphal (1998) warn against unreliable estimates from PVA’s and suggest that PVA’s are used to evaluate relative rather than absolute rates of extinction, emphasize short-time periods, examine feasible scenarios and that they are simple models. Few PVA’s have been done on plants (Menges, 2000). Two South African examples are Pfab and Witkowski (2000) and Raimondo and Donaldson (2003).. Matrix models have been used to determine general population dynamics and demography of population in plants.. Some studies have determined the effect of disturbance on. populations. Examples of studies on the effect of harvesting of exploited resources of natural plant populations using matrix models include Desmet et al. (1996) and Lamont et al. (2001). A study by Maze and Bond (1996) used a simple static model that dealt with harvesting of proteas in the fynbos. Studies have also been conducted on the harvesting of wildlife, for example Jensen (1996) and Loon and Polakow (1997). Grazing has a similar effect on populations as harvesting and acts as a disturbance.. Matrix models have been used to. determine the effect of this disturbance. These include Thalen et al. (1987), Bullock et al. (1994) and Ehrlén (1995). Fire is another disturbance in plant populations and examples of. 14.

(29) studies using population projection matrices to determine the effects of fire include Canales et al. (1994), Gross et al. (1998) and Menges and Dolan (1998).. 1.3.4. Model testing. Model validation should be an important section of all modelling papers (Aber, 1997). Model validation concerns our degree of faith in the quality of the model with respect to reality (Haefner, 1996).. This is particularly important when models are used to solve. management problems (Bart, 1995). The concept of quality is linked to the uses and aims of the model but can include usefulness for management, understanding or insight provided, accuracy of predictions, simplicity, generality, robustness and cost (Haefner, 1996). Models for understanding are evaluated by their accuracy and the range of conditions over which they are useful (Caswell, 1971). A model is a closed system but real systems are open which allows for changes in their environments (Hannon and Ruth, 1997). It is not possible to completely verify a model against reality because there may be extreme situations where the real system behaves differently than the model (Oreskes et al., 1994, Hannon and Ruth, 1997). A modeller aims to capture the essentials of the real system at the expense of other factors and thus model verification deals with consistency or logical accuracy (Hannon and Ruth, 1997). Model reliability can be tested for in the model structure, parameter values, secondary predictions and primary predictions (Bart, 1995).. Models can be tested by withholding some of the basic data used in model construction or by predicting the condition in some unmeasured area and then measuring these variables in the field (Hannon and Ruth, 1997). If it is not possible to verify the model by comparing its results to a real system then we may not know whether we captured the essentials of the. 15.

(30) system (Hannon and Ruth, 1997). If the results of a simple model are unable to reproduce the empirical observations then something is wrong with the model and revision is necessary (Hannon and Ruth, 1997). If the results coincide with observations of the real system then the possibility still exists that the errors in the model cancel each other (Hannon and Ruth, 1997).. Different views are held on the importance of accuracy: Starfield (1997) believes that the test of a model is not its accuracy but whether it aids decision-making. If a model is viewed as a problem-solving tool then validation is irrelevant and justification of assumptions and sensible interpretation is important (Starfield, 1997). Models are representations that are useful for guiding but not susceptible to proof (Orekes et al., 1994). Despite the problem of verification of models, they remain useful as they can be used for sensitivity analyses and answer ‘what if’ questions (Orekes et al., 1994)and are thus appropriate for this study.. 1.4. Literature Review: The Fynbos Biome. The Fynbos Biome occupies 2.7% (71 337 km2, in the southwestern and southern parts) of South Africa (Cowling et al., 1997). Rainfall occurs in the winter months and is typically Mediterranean in the west, and changes to non-seasonal in the southeast (Cowling et al., 1997). Fynbos (‘fine leaved bush’) is structurally characterised by six functional types, these are ericoid shrubs, fire ephemerals, restioids, geophytes, obligate resprouting shrubs and proteoid shrubs (Cowling et al., 1997). Furthermore, it is evergreen, fire-prone shrubland that is largely confined to sandy, infertile soils (Cowling et al., 1997). Soils tend to have low pH values, low available phosphorous, low base status and high clay content (Cowling et al., 1997).. 16.

(31) The Fynbos biome includes about 7300 species (Cowling et al., 1997) and approximately 80% of these species are endemic (Cowling et al., 1997). Roughly 68% of the species in the Cape Floristic Region (CFR) are endemic (Cowling and Holmes, 1992). The CFR is one of the world’s six floristic kingdoms and roughly coincident with the Fynbos biome (Cowling et al., 1997).. Further, the CFR has the highest recorded species density, compared to. equivalent-sized areas, for any temperate or tropical region in the world (Cowling and Holmes, 1992).. The Fynbos biome is distinguished from other South African biomes by the presence of a large number of species in the Ericaceae, Restionaceae, Rutaceae, Polygalaceae, Thymelaceae, Rhamnaceae, Rosaceae and Lobeliaceae (Gibbs Russell, 1987). The Fynbos is characterised by a co-dominance of hemicryptophyte, chamaephyte and phanerophyte life forms (Cowling et al., 1997). Fynbos is differentiated from other biomes by climate, nutrientpoor soils and recurrent, intense summer fires (Cowling et al., 1997).. Cowling and Holmes (1992) found considerable differences between the regional floras in the southwestern and southeastern floras of the CFR. The trends that they observed (Cowling and Holmes, 1992) were a concentration of taxa (including the Restionaceae) in the extreme southwest; higher species-to-genus ratios and proportion of species in the southwestern floras; and a lower frequency of endemic species in the southeastern flora.. The most logical. explanation is a historical one linked to differences in paleoclimatic conditions where the southeastern flora were limited to refugia (Cowling and Holmes, 1992).. This variation in ‘fynbos’ across the biome has resulted in many definitions of Fynbos (see Cowling and Holmes, 1992 for a brief discussion on the history of these definitions). The. 17.

(32) currently accepted definition (Campbell, 1986) essentially states that the presence of restioids is the essential feature of Fynbos. Over 60% of the plots sampled by Campbell had >30% restioid cover thus his conclusion that they are important in defining Fynbos (Cowling and Holmes, 1992).. The functional types are distinguished based on their structure and reproductive ecology. Many Fynbos reproductive traits (e.g. canopy-stored seeds, the abundance of fire-killed shrubs) are rare in other biomes (Le Maitre and Midgley, 1992). Reasons given for the evolution of these rare reproductive traits are the limitation of soil nutrients, the variable rainfall regime and fire (Le Maitre and Midgley, 1992). Soil nutrients influence the cost of reproductive structures, such as elaiosomes (fatty deposits on seeds that attract ants for dispersal, Cowling et al., 1997), which are inexpensive in terms of nutrients, or dormant buds, which have a low nutrient cost despite the carbohydrate requirement; these options are low in cost compared to nutrient allocation to seeds (Le Maitre and Midgley, 1992). Rainfall is often a cue for seed germination and growth of individuals (Le Maitre and Midgley, 1992) and may limit flowering slightly (Cowling et al., 1997). Fire stimulates seed release, seed germination and flowering in a variety of species (Le Maitre and Midgley, 1992). These three factors can account for most of the variation in reproductive traits.. 1.5. Literature Review: Economic uses of Fynbos. Fynbos ecosystems provide a wide range of valuable ecosystem services including the thatching industry, such as ecotourism opportunities, biodiversity storage and water production (Cowling and Costanza, 1997).. A study by Turpie et al. (2003) aimed at. determining the economic value of both terrestrial and marine biodiversity in the Cape. 18.

(33) Floristic Region (CFR). The study was done because individuals will only conserve if incentives exist; and if the costs of conservation are outweighed by the benefits (Turpie et al., 2003). This is relevant in the fynbos as the priority conservation areas that have been identified are so large that some protection is due to off-reserve conservation measures (Turpie et al., 2003). The estimation of biodiversity value in monetary terms helps promote and justify conservation actions (Turpie et al., 2003). Values can range from consumptive values through to existence values (Turpie et al., 2003).. Consumptive values from fynbos products include foods (such as sour figs (Carpobrotus edulis) and honeybush tea (Cyclopia intermedia), Buchu (Agathosma spp.), thatching reed and wildflowers (Turpie et al., 2003). The thatching industry is about R15.5 million per year (Cowling and Richardson, 1995), which translates to an average added value at farm gate level nett to farmerof R49.46 per hectare in Limestone Fynbos. The thatching industry has a greater added value than the other consumptive uses from the Fynbos; the one exception is the wildflower industry. The wildflower industry is comprised of approximately 100 species (Cowling and Richardson, 1995) of flowers (e.g. Protea spp.) and greens (e.g foliage, Erica spp.) for the fresh industry and cones and other flowers for the dried flower industry (Turpie et al., 2003). The value added per hectare at farm gate level for Limestone Fynbos from the wildflower industry is R55.61, compared to R49.46 for thatch (Turpie et al., 2003).. The wildflower industry provides employment in areas otherwise unsuitable for agriculture, owing to low soil fertility and rainfall conditions (Cowling et al., 1997). However, it is important that harvesting is practised sustainably for both economic and conservation reasons (Cowling et al., 1997). An example of sustainable harvesting is Flower Valley on the Agulhas plain (Privett, 2002). Sustainable production, while minimising the. 19.

(34) impact on community structure and process is important; over-harvesting a plant so that it becomes extinct is unacceptable, as is a management practice which results in sustainable production of a target species but the extinction of another species in the community (Cowling, 1989).. The sustainable use of biodiversity to protect ecosystem services,. improving harvesting techniques of resources and promoting sustainable nature-based tourism is one of the three broad goals of the Cape Action Plan for the Environment (Gelderblom et al., 2003).. The impacts of the wildflower and thatching industry extend beyond the direct effects on target species (Manders, 1989). These include: direct-abuse (due to a lack of interest in longterm sustainability or lack of knowledge of threatened species); soil erosion (due to roads built for access of populations); harvesting practices (plants damage, loss of seeds due to over-picking and trampling); fungicides and pesticides (destroy pollinator populations); and orchard cultivation of wildflowers (cross pollination with natural populations, transfer of diseases and pests and influences on pollinators) (Manders, 1989). There is also the risk of negatively impacting tourism if harvesting results in bare and unattractive veld because natural or semi-natural attractions are the primary reason that 80% of tourists visit the CFR (Turpie et al., 2003). The overall contribution of nature based tourism is R7443 million or 7.2% of the Western Cape Gross Geographic Product (provincial contribution to the GDP) (Turpie et al., 2003).. Indirect uses of the Fynbos include ecological functions that facilitate other industries such as bee-keeping and fruit production (Turpie et al., 2003). The mentioned values are all use values and only existence values are considered non-use values (Turpie et al., 2003).. The. total economic value of the CFR’s natural resources, about R10,000 million per year,. 20.

(35) represents over 10% of the Western Cape’s Gross Geographic Product (Turpie et al., 2003). Higgins et al. (1997) did a similar study to Turpie et al. (2003) using estimates rather than actual data and showed similar results. An appreciation of the full economic value helps protect areas, as the CFR’s natural systems’ economic value is significant (Turpie et al., 2003). The economic importance of these Thamnochortus species indirectly acts as a draw card for conservation, as quantification and appreciation by society of the services provided by a natural ecosystem provides a powerful incentive for their conservation (Cowling and Costanza et al., 1997). Management that ensures the sustainability of natural populations of thatching reed is important.. 1.6. Literature Review: Fire. 1.6.1 Fire and life-histories. Complex responses to fire occur in the Fynbos Biome, these are a result of variation in the fire regime combined with the diverse range of fire life histories (Bond, 1997). Many Fynbos species are killed by fire because they have no capacity to sprout and are thin barked (Bond, 1997). These species, non-sprouters, make up 50% of the species in Fynbos communities (Le Maitre and Midgley, 1992). Many Restionaceae species and the dominant woody shrubs in Fynbos are non-sprouters (Bond, 1997). Species that survive fire by sprouting show variation in their ability to sprout because of the location of sprouting tissue and the extent to which sprouting occurs among sizes, species and after different fires (Bond, 1997). Some plants have no sprouting capacity but survive if shoot apices are undamaged while others have protected bud banks from which the canopy regenerates (Bond, 1997). Other species have basal sprouting or lignotubers or root suckers that facilitate sprouting (Bond, 1997).. 21.

(36) Furthermore, sprouting includes a continuum of behaviour with species ranging from those that sprout only after light disturbances to more robust species; this variation is dependent on what organ produces new shoots (Midgley, 1996). Sprouting also varies among life history stages and disturbances of differing severity (Bond and Midgley, 2001), is not limited to periods of disturbance and can occur in the absence of disturbance (Midgley, 1996).. Classifications are useful for generalising population responses to fire because of the above-mentioned variation in response to fire. Consideration of how the plant survives, and reproductive responses to fire produces four life history strategies (Bond, 1997). A plant is classified as being either fire-recruiting or non-fire-recruiting and either sprouting or nonsprouting (Bond, 1997). These life histories are a result of evolutionary selection pressures, occur across different parts of plant communities and occur at different stages in post-fire succession (Bond, 1997).. Long-lived non-sprouting species are common in Fynbos with long (>10 years) fire intervals, as are fire-recruiting species (Bond, 1997). The woody species of Fynbos are divided into three regeneration syndromes or life histories: non-sprouting, facultative sprouting, and obligate sprouting (Le Maitre, 1992). Fire-recruiting non-sprouters are killed by fire, accumulate seed banks between fires, and regenerate only after fires. Facultative sprouters regenerate after fires from both dormant buds and seeds accumulated between fires, and may experience high mortality rates in intense fires while obligate sprouters produce few seedlings (almost exclusively between fires), survive fires by sprouting and have bird dispersed seeds (Le Maitre, 1992). Population growth in non-fire-recruiting sprouters (plants that sprout and are not limited to recruitment in post-fire conditions/ facultative sprouters) is such that populations grow between fires and population growth is continuous because of. 22.

(37) seedling recruitment and fires increase juvenile and adult mortality (Bond and van Wilgen, 1995).. The non-fire-recruiting non-sprouters are rare in fire-prone vegetation and are. normally short-lived species with well-dispersed seeds (Bond and Van Wilgen, 1995). Firerecruiting sprouters (obligate sprouters) have discrete population growth with multiple cohorts from previous fires; populations generally decline between fires (Bond and Van Wilgen, 1995).. Fire-recruiting non-sprouters form even-aged cohorts with non-overlapping. generations from the last burn and numbers decline between fires; in this group, the seed bank is important as future populations rely on seeds (Bond and Van Wilgen, 1995). It is important to understand these life histories in order to understand population responses to burning (Bond, 1997). Thamnochortus insignis relies on seeds for population regeneration but does not rely entirely on fire for recruitment and is thus a non-fire-recruiting non-sprouter (Ball, 1995). T. erectus also does not rely on fire for regeneration and can be classified as a facultative sprouter or non-fire-recruiting sprouter (Ball, 1995).. 1.6.2. Components of Fire. At present, the general fire management strategy in the Fynbos Biome is based on three components of fire: frequency, season and intensity (van Wilgen et al., 1992). Fire frequency is dependent on the fuel load (which can vary inter-annually) and thus has physical limitations (van Wilgen et al., 1992). Plant communities will respond differently to the frequency of fire based on their vital attributes (van Wilgen et al., 1992). Prolonged absence of fire may lead to cyclic succession due to population senescence or death of large individuals (Kruger, 1983). The following component that influences vegetation structure is fire seasonality. Seasonal variation in fuel sources is not a feature of Fynbos vegetation thus fire seasonality is determined by climatic factors (van Wilgen et al., 1992). Fires occur mainly in the summer. 23.

(38) (November to March) but can occur in all months under suitable conditions (van Wilgen, 1987). Fire intensity is a measure of how fiercely a fire burns (Bond, 1997), which depends on fuel loads and the rate at which they burn (van Wilgen et al., 1992). Fynbos fires are generally crown fires that consume all layers of the vegetation (Bond, 1997). The intensity can be varied by reducing fuel loads or by changing fire conditions (van Wilgen et al., 1992). Fire intensity is important to species with soil-stored seeds (Bond, 1997) through heat stimulation of germination or lethal exposure. The effects of a burn on vegetation dynamics depend on the conditions at the time of the event but also on the population structure immediately before the burn (Bond, 1997). Various combinations of these three components of fire and population structure can result in varied plant communities. Fire can change the structure of the vegetation, namely the balance between sprouters and non-sprouters (Kruger, 1983). For example a high intensity fire can have negative effects on sprouting species while non-sprouting species are more sensitive to changes in fire frequency (van Wilgen et al., 1992); relative abundances of non-sprouting species are also affected by season of burn with successful regeneration after summer or autumn fires (Kruger, 1983). This is because of the interaction between fire-regime and species traits (regenerative properties) (Bond, 1997). Sprouters tend to be more resilient under a range of fire regimes than non-sprouters, and show remarkable powers of vegetative recovery (Smith et al., 1992).. Thus, fire frequency,. seasonality and intensity influence vegetation structure and determine vegetation dynamics.. 1.6.3. Non-sprouter: Sprouter balance. Fire provides an explanation for the high α diversity of Fynbos and partially explains why sprouters may be favoured in the Fynbos due to too-frequent fires. However, such obvious fire effects do not explain the considerable variation in the relative abundances of sprouting. 24.

(39) and seeding shrubs (Kruger, 1983). Two explanations have been given for why non-sprouters or sprouters dominate. Keeley (1977, from Kruger, 1983) argues, from an American context, that the availability of space is what determines non-sprouter success (Fire frequency hypothesis , Bond and Van Wilgen, 1995); if space is available for non-sprouter seedlings to grow free from competition then non-sprouters can dominate. Such a situation can occur if fire is excluded for a sufficiently long time that the sprouters begin to self thin and fuel accumulates so that sprouter mortality is increased after fire. This is in comparison to a seedling trying to grow in a frequently burnt area where there is dense sprouter re-growth. In conclusion, Keeley (1977, from Kruger, 1983) argues that the non-sprouter: sprouter ratio will increase as the fire frequency decreases and thus a variable fire frequency is required for coexistence of both species. The Gap-availability hypothesis (Bond and Van Wilgen, 1995) of Specht (1981) provides an alternative explanation from an Australian context. Specht also emphasizes the competitive advantage of sprouters in the post-fire environment but he differs from Keeley in that he states that dominance of either life-history is reliant on site productivity rather than fire history. Specht’s argument deals with available space, like Keeley’s argument, as he states that available moisture determines the density of foliage; thus the more humid the environment, the fewer gaps available for non-sprouters.. From a South African Fynbos context, Kruger (1983) states that seeding shrubs may cover half or more of the ground and sprouters are not as dense; but this argument is inaccurate because most species recruit after fire when the canopy is severely reduced (Smith et al., 1992). Smith et al., (1992) found at Swartboskloof, that the relative importance of nonsprouters and sprouters in the mesic mountain fynbos changes along a soil moisture gradient; furthermore, non-sprouters were more prevalent in xeric sites, which supports Specht’s (1981) hypothesis. Senescence in seeding species results in reduced seed banks and a decline in. 25.

(40) populations after fire thus the non-sprouter: sprouter balance responds strongly to both an increase or decrease in fire frequency (Kruger, 1983). Midgley (1996) predicts (based on South African data) that because non-sprouters are taller than sprouters that sprouters will dominate shorter or sparser vegetation and thus the more stressful parts of biomes. He further predicts that, within a genus, not all non-sprouters will be tall but that the tallest species will be a non-sprouter. According to Le Maitre and Midgley (1992), sprouting did not only evolve in response to fire, and it may be a primitive or a derived trait depending on the taxon. Despite this the divergent patterns of traits between sprouters and non-sprouters suggests that they are different life-histories rather than differences from a single trait and thus will respond differently to disturbances.. 1.7. Literature Review: Life-history strategies. Non-sprouters differ from sprouters because of a division of resources that differs between the two life-history forms and trade-offs occur between survival and reproduction. The constraining relationships and trade-offs between demographic traits determine individual fitness (Stearns, 1992). Life history evolution looks at why and how organisms have evolved different ways of combining traits to affect fitness (Stearns, 1992). The traits that form parts of a life history are: size at birth, growth pattern, age at maturity, size at maturity, number size and sex ratio of offspring, age and size-specific mortality schedules and life span (Stearns, 1992). Trade-offs occur between these traits, the most commonly observed being those between: current reproduction and survival, current reproduction and future reproduction and the number, size and sex of offspring (Stearns, 1992).. 26.

(41) The core of life-history theory is the concept of allocation (Bond and Van Wilgen, 1995). Resources should be allocated in such a way that fitness is maximised, and sprouters and nonsprouters have different allocation patterns (Bond and Van Wilgen, 1995). Among sprouters there is a trade-off between the certainty of persistence by sprouting and regeneration by seedling establishment (Bond, 1997). Sprouters allocate resources, to ensure persistence or fire survival, to storage organs while non-sprouters shed these costs and gain reproductive and growth rate advantages. Sprouters offset the risk of recruitment failure associated with complete seed germination by the survival and persistence of established plants; nonsprouters reduce the risk by spreading germination over several episodes (Le Maitre, 1992). The generalisation is that plants that are vigorous sprouters (as adults) tend to have low recruitment compared to non-sprouters that recruit more readily (Bond and Midgley, 2001). In Mediterranean shrublands, it has been observed that ‘sprouters have fewer seeds, smaller seed banks, slower growth and maturation rates and almost always have fewer seedlings and poorer seeding survival than non-sprouters’ (Bond and Midgley, 2001). Studies of shrubs in Chaparral show that sprouter ability to survive fires is negatively correlated with seed production and seedling establishment (Le Maitre, 1992).. As previously mentioned, sprouters are also generally shorter than non-sprouters (Midgley, 1996). Hodgkinson (1998) found that sprouters (species he defined as relying more on sprouting than recruitment for population persistence) maintained maximum sprouting success with shrub height growth. In comparison, the non-sprouters (species relying largely on recruitment from seed to maintain populations) were not able to sprout once seedlings were established or lost the ability to maintain sprouts. Dependence on obligate reseeding in unproductive habitats is risky because individuals’ growth is slower and plants are more likely to be killed by disturbance before recruiting seedlings (Bond and Midgley, 2001). The. 27.

(42) generality of the trade-offs between sprouting and seeding are unknown (Bond and Midgley, 2001).. Other life-history traits are associated with regeneration strategy (Verdú, 2000). Pate et al. (1990) found that sprouters have slower shoot growth, lignotubers with less extensive root systems and delayed flowering/ maturation. It has also been found that seedlings of nonsprouters grow significantly faster than those of sprouters (Verdú, 2000). Further, Verdú (2000) observed that non-sprouters are mainly non-sclerophyllous, dry fruited, small seeded species associated with early colonisation; in comparison to this, sprouters are mainly sclerophyllous, vertebrate dispersed, fleshy fruited, large-seeded species associated with late succession.. Pate et al. (1990) studied the patterns of growth and storage of minerals and starch in seedlings (2- 4 years old) of a range of non-sprouter and sprouter species, commonly found in southwestern Australia, that were recruiting together from seed in habitats recently exposed to fire. Their results show that non-sprouting species allocated four times more assimilates to the above ground parts than the sprouters; this was seen in a greater shoot: root dry weight ratio for non-sprouter species (Pate et al., 1990). The non-sprouters also had a greater total plant dry weight, while the sprouters had more starch in root dry matter but both species exhibited greater starch storage in roots than shoots (Pate et al., 1990). According to Pate et al. (1990), fire sensitive non-sprouter species are clearly equipped for maximising seed production quickly due to fast shoot growth and early attainment of mature stature and reproductive capacity, and thereby establish an effective seed bank before the next fire. In comparison to this, sprouters are equally well adapted to survive fire through their development of below ground biomass in the form of tap roots or lignotubers, from which. 28.

(43) new shoots readily arise once the shoot system has been destroyed (Pate et al., 1990). Hansen et al. (1991) did a similar study on shrubby legumes (Bossiaea and Fabaceae) and showed similar results to Pate et al. (1990) for the non-sprouter and sprouter species. They also observed that the non-sprouters exhibit a high rate of juvenile growth and attain maximum reproductive output early while the sprouters grow slowly from seed but have a higher rate of shoot growth after fire than the non-sprouters (Hansen et al., 1991). Bell and Ojeda (1999) studied underground starch storage in sprouting and seeding Erica species in the Cape Floristic Region and found similar patterns of root starch allocation to Pate et al. (1990) and Hansen et al. (1991); however, the shoot starch did not show different concentrations between non-sprouters and sprouters (Bell and Ojeda, 1999). They also found that the sprouting species have a larger amount of specialised storage tissue (Bell and Ojeda, 1999). Sprouting species would be expected to have greater resources of starch in underground organs, in order to re-grow following above-ground damage, than non-sprouters (Bell and Ojeda, 1999).. In a study on the contrasting growth and morphological characteristics of non-sprouter and sprouter species of Restionaceae from southwestern Australia, Pate et al. (1991) observed that the culm: rhizome dry weight ratio was greater in non-sprouters than sprouters, and that the sprouter’s perennating buds were more deeply buried. Starch was present at detectable levels in 23 sprouter species and only 3 non-sprouter species (Pate et al., 1991). The level of total soluble sugars was also lower in the non-sprouter species (Pate et al., 1991). Despite these correlations with the previous studies on other families, it is unknown whether the seedling growth rate, timing of reproduction and reproductive effort patterns observed in the previous studies are applicable to the Restionaceae (Pate et al., 1991).. 29.

(44) In a study by Meney et al. (1997), an attempt was made to determine if these other patterns are applicable to herbaceous perennial monocots (including the Restionaceae) in southwest Western Australia. Their results showed that more seeds were produced per ovule, which results in greater pre-dispersal reproductive success, by fast growing, shallow rooted nonsprouters than in slow growing, long-lived sprouters (Meney et al., 1997). In a strongly clonal species (Alexgeorgea subterranea), the seed: ovule ratio is much lower than in other non-sprouters which suggests that reproductive success was compromised by diversion of resources to vegetative spread; this is expected in clonal sprouter species (e.g. A. nitens) which will be reproductively disadvantaged by investing resources into thick, deeply buried rhizomes with carbohydrate reserves (Meney at al, 1997). A further comparison between congeneric sprouter and non-sprouter pairs with similar phenologies and environmental conditions shows that fruit and seed set was higher in the non-sprouter and thus reproductive success is partially determined by resource allocation associated with regeneration mode (Meney et al., 1997). There was no clear relationship between post-dispersal success and regeneration mode (Meney et al., 1997). The assessments of fertile seeds per year per female plant (fecundity) show that all species in Meney et al.’s (1997) study (except two) are potentially capable of replacing the population annually. Parent: seedling ratios of almost all species (except 3 fecund non-sprouter species) are estimated to be less that eight thus species resilience to stress is unknown (Meney et al., 1997). The number of fertile seeds that survive until germination and the number of these seedlings that survive until reproduction affect the number of reproductive adults in the next generation (Meney et al., 1997). The combination of low pre-dispersal reproductive success and post-dispersal stresses will restrict the redevelopment of populations from seed (Meney et al., 1997). The maintenance of a seed bank is important in order to guarantee population persistence after a stress or disturbance and. 30.

No results found